Pref-GUIDE: Continual Policy Learning from Real-Time Human Feedback via Preference-Based Learning

TL;DR

Pref-GUIDE introduces a novel method to convert real-time scalar human feedback into preference data, enhancing reward learning and policy training in online reinforcement learning scenarios, especially under noisy feedback conditions.

Contribution

It proposes a preference-based framework that improves reward model accuracy and robustness by filtering and aggregating human feedback, outperforming scalar-feedback baselines.

Findings

Pref-GUIDE outperforms scalar-feedback methods in three environments.

Voting variant surpasses dense reward performance.

Framework effectively handles noisy and inconsistent feedback.

Abstract

Training reinforcement learning agents with human feedback is crucial when task objectives are difficult to specify through dense reward functions. While prior methods rely on offline trajectory comparisons to elicit human preferences, such data is unavailable in online learning scenarios where agents must adapt on the fly. Recent approaches address this by collecting real-time scalar feedback to guide agent behavior and train reward models for continued learning after human feedback becomes unavailable. However, scalar feedback is often noisy and inconsistent, limiting the accuracy and generalization of learned rewards. We propose Pref-GUIDE, a framework that transforms real-time scalar feedback into preference-based data to improve reward model learning for continual policy training. Pref-GUIDE Individual mitigates temporal inconsistency by comparing agent behaviors within short windows and filtering ambiguous feedback. Pref-GUIDE Voting further enhances robustness by aggregating reward models across a population of users to form consensus preferences. Across three challenging environments, Pref-GUIDE significantly outperforms scalar-feedback baselines, with the voting variant exceeding even expert-designed dense rewards. By reframing scalar feedback as structured preferences with population feedback, Pref-GUIDE offers a scalable and principled approach for harnessing human input in online reinforcement learning.